Korean J Ophthalmol 2017;31(6):538-547 https://doi.org/10.3341/kjo.2017.0017 pISSN: 1011-8942 eISSN: 2092-9382

Original Article Efficacy and Safety of Intracameral Bevacizumab for Treatment of Neovascular

Jun Young Ha1, Tae Hee Lee1, Mi Sun Sung1, Sang Woo Park1,2

1Department of Ophthalmology and Research Institute of Medical Sciences, Chonnam National University Hospital, Chonnam National University Medical School, Gwangju, Korea 2Center for Creative Biomedical Scientists, Chonnam National University, Gwangju, Korea

Purpose: To evaluate the long-term efficacy and safety of intracameral bevacizumab in patients with neovas- cular glaucoma. Methods: This retrospective study included 26 eyes of 26 neovascular glaucoma patients who received in- tracameral bevacizumab injection between January 2013 and May 2015, and were followed-up for at least 1 year. All patients were treated with topical and/or systemic intraocular pressure (IOP)-lowering medica- tions, intracameral bevacizumab, and panretinal photocoagulation (PRP). The main outcome measures were changes in visual acuity, IOP, and neovascularization of the (NVI) and the anterior chamber angle (NVA). To assess the safety of intracameral bevacizumab, corneal endothelial changes were also determined using specular microscopy. Patients whose IOP was uncontrolled received IOP-lowering surgery. Clinical factors associated with IOP-lowering surgery were also investigated. Results: In all patients, intracameral bevacizumab resulted in a rapid and marked reduction of IOP, NVI, and NVA within 1 week. At 12 months after initial injection, 19 of 26 eyes (73%) underwent IOP-lowering surgery. The average interval between initial injection and surgical treatment was 33.6 ± 26.9 days. Baseline IOP (p = 0.018), NVA grade (p = 0.029), and incomplete PRP (p = 0.005) were identified as predictive factors for IOP-lowering surgery. During the follow-up period, there were no statistically significant corneal endothelial changes after intracameral bevacizumab injection. Conclusions: During 1 year of follow-up after intracameral bevacizumab, the procedure was found to be safe for the corneal endothelium. However, the IOP-lowering effect was transient, and 73% of patients eventually required IOP-lowering surgery. Predictive factors for IOP-lowering surgery were high baseline IOP and NVA grade, and incomplete PRP.

Key Words: Bevacizumab, Intracameral injection, Neovascular glaucoma, Vascular endothelial growth factor

Neovascular glaucoma (NVG) is a secondary glaucoma, such as proliferative diabetic (PDR), retinal resulting from ocular ischemia caused by various diseases vein occlusion, , and chronic [1]. Ischemia-induced abnormal fibrovascular tissue grows on the iris and anterior chamber angle and progress- Received: January 31, 2017 Accepted: May 10, 2017 es to produce obstruction of the trabecular meshwork, Pe- Corresponding Author: Sang Woo Park, MD, PhD. Department of Oph- ripheral anterior synechiae (PAS), and angle closure. These thalmology, Chonnam National University Hospital, #42 Jebong-ro, Dong-gu, Gwangju 61469, Korea. Tel: 82-62-220-6742, Fax: 82-62-227- histopathological changes eventually result in elevated in- 1642, E-mail: [email protected] traocular pressure (IOP) and severe vision loss.

© 2017 The Korean Ophthalmological Society This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses /by-nc/3.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

538 JY Ha, et al. Intracameral Bevacizumab in Neovascular Glaucoma

There are two goals for management of NVG: 1) treat of 14 eyes (85.7%) required additional laser or surgical pro- the underlying condition responsible for the neovascular cedures 6 months after intravitreal bevacizumab. stimulus, and 2) control the elevated IOP, which results in Although the rapid and obvious therapeutic effects of in- progressive if untreated. Panretinal pho- traocular bevacizumab have been reported in many stud- tocoagulation (PRP) is the standard treatment for ablation ies, little is known about the long-term efficacy of intraoc- of the ischemic and induces involution of the vasop- ular, especially intracameral, injection. In addition, roliferative pathology that includes abnormal blood vessel previous studies have reported that most patients eventual- growth on the retina, iris, and anterior chamber angle. ly required laser or surgical procedures for IOP control, However, the manifestation of the anti-neovascular effect but there have been very few studies on the predictive fac- driven by PRP requires several weeks, and PRP usually tors for surgical treatment in spite of intraocular injection, does not induce prompt regression of iris and angle neo- especially intracameral injection [12]. Therefore, the pur- vascularization. Therefore, ocular damage may persist due pose of the present study was to observe the long-term ef- to elevated IOP in the eyes of NVG patients [2,3]. More- ficacy and safety of intracameral bevacizumab and to in- over, adequate laser treatment is often difficult due to me- vestigate the clinical factors associated with IOP-lowering dia opacity caused by corneal , , , surgery, despite the use of intracameral bevacizumab in and/or vitreous hemorrhage. Vascular endothelial growth NVG patients. factor (VEGF), induced by ocular ischemia, and plays a central role in ocular neovascularization and NVG [4,5]. Aqueous levels of VEGF are highly elevated and signifi- Materials and Methods cantly correlated with the extent of neovascularization and IOP in NVG patients [6,7]. Based on clinical and experi- This study was approved by the institutional review mental findings, anti-VEGF therapy was proposed as an- board of Chonnam National University Hospital (CNUH- other therapeutic option in NVG [8]. 2017-017) and is in accordance with the principles of the Bevacizumab (Avastin; Genentech, South San Francisco, Declaration of Helsinki. The authors retrospectively re- CA, USA), a humanized monoclonal antibody against all viewed the medical records of 26 eyes of 26 patients who VEGF isoforms, is widely used to treat VEGF-mediated underwent intracameral bevacizumab treatment for NVG ocular conditions, such as choroidal neovascularization, between January 2013 and May 2015. age-related , and diabetic macular NVG was defined as IOP >21 mmHg with the presence edema. Bevacizumab is also used as an adjunctive agent in of NVA or NVI. Patients previously treated with intraocu- glaucoma surgery. Previous studies have reported that in- lar anti-VEGF injection or anti-glaucoma surgery, and traocular bevacizumab injection can cause a rapid and those with severe media opacity preventing examination of marked regression of neovascularization of the iris (NVI) the anterior segment were excluded. and the anterior chamber angle (NVA), and can reduce All subjects received anti-glaucoma topical and/or sys- IOP in early-stage NVG. Wolf et al. [9] reported complete temic medications, intracameral bevacizumab, and PRP. resolution of NVI and IOP reduction in all patients 1 week PRP was performed with a slit lamp using a 532-nm green after intracameral bevacizumab. Sasamoto et al. [6] found laser (OcuLight GL; Iridex, Mountain View, CA, USA). A a significant reduction of NVI, NVA, and IOP 1 week after level II to III reaction for retinal photocoagulation was ap- intravitreal bevacizumab. propriate for laser output power intensity, the spot size was However, the effect of intraocular bevacizumab for 300 to 500 microns, the exposure time was 0.1 to 0.2 sec- NVG may be transient. The half-life of intravitreal bevaci- onds, and the photocoagulation scope was 0.5 to 1.0 papilla zumab was reported to be only 9.8 days in humans and diameter (PD) on the nasal side in the and 2 PD 4.32 days in animals [9,10]. Wolf et al. [9] demonstrated on the temporal side in the macula and the left retina out- that the effect of intracameral bevacizumab lasted approx- side of both the up and down vascular arcades. Depending imately 23 days, and 6 months after injection, 21 of 24 on the view of the retina and patient tolerance, PRP was (87.5%) eyes required additional treatment, such as repeat- done over 1 to 3 sessions. Supplemental PRP was added to ed injection or PRP. Yazdani et al. [11] also reported that 12 areas of nonperfusion revealed by fundus fluorescein angi-

539 Korean J Ophthalmol Vol.31, No.6, 2017

ography (FFA) at 1 to 2 months after the first PRP. We de- face neovascularization of the pupillary zone of the iris, fined complete PRP as all of the NVI and NVA were re- involving ≤2 quadrants; grade 2, surface neovasculariza- gressed after PRP or all of the nonperfusion area was tion of the pupillary zone of the iris, involving ≥2 quad- laser-ablated so there was no need for additional photoco- rants; grade 3, in addition to the pupillary zone, neovascu- agulation. larization of the ciliary zone of the iris and/or During follow-up, intracameral bevacizumab was re- uveae involving 1 to 3 quadrants; grade 4, neovasculariza- peated if IOP had increased to >21 mmHg despite medical tion of the ciliary zone of the iris and/or ectropion uveae and laser treatment, or if there was a prominent recurrence involving ≥3 quadrants. NVA grading is as follows: grade of NVI and/or NVA. Despite all efforts to control IOP, 1, fine neovascular twigs cross scleral spur and ramify on IOP-lowering surgery, such as trabeculectomy with mito- trabecular meshwork, involving ≤2 quadrants; grade 2, mycin C, Ahmed valve implantation, or trans scleral cyclo- neovascular twigs cross scleral spur and ramify on trabec- photocoagulation, was performed by a single glaucoma ular meshwork, involving ≥2 quadrants; grade 3, in addi- specialist (SWP) when IOP was above the target with pro- tion to the trabecular meshwork, PAS involving 1 to 3 gression of glaucomatous optic neuropathy. The target IOP quadrants; grade 4, PAS involving ≥3 quadrants. was estimated to prevent further nerve damage and was All patients were divided into two groups, patients who set for each patient based on their initial IOP and degree of received IOP-lowering surgery and patients whose IOP existing damage. was controlled without surgery. The changes in BCVA, Intracameral bevacizumab injection was performed at IOP, NVI, and NVA were acquired and compared between an outpatient clinic. Informed consent was obtained from the two groups. each patient prior to the procedure. After aseptic prepara- The frequency of IOP-lowering surgeries during follow- tion (5% povidone-iodine solution) and application of topi- up, the number of injections, the number of patients re- cal anesthetic eye drops (proparacaine hydrochloride 0.5%; ceiving multiple injections, the interval between initial in- Alcaine, Alcon, Fort Worth, TX, USA), bevacizumab solu- jection and surgery, the type of surgery, and predictive tion (25 mg/mL, 0.05 mL) was injected at the limbus in the factors for surgery were also assessed. In addition, endo- temporal quadrant, using a 30-gauge needle after paracen- thelial cell counts, coefficient of variation, and hexagonali- tesis. Before injection, paracentesis (0.1 to 0.2 mL) was ty were evaluated via specular microscopy to assess corne- performed to prevent IOP elevation due to the injected so- al endothelial toxicity due to the injected bevacizumab. lution [12]. Statistical analysis was performed using PASW Statis- All subjects underwent comprehensive ophthalmologic tics ver. 18.0 (SPSS Inc., Chicago, IL, USA). Pre-injection examinations, including detailed medical histories, mea- baseline BCVA, IOP, NVI, and NVA were compared with surements of best-corrected visual acuity (BCVA) by loga- post-injection follow-up examination values using paired t- rithm of the minimum angle of resolution (logMAR) scale, tests and the paired McNemar test. The Mann-Whitney U- anterior segment and fundus examinations with slit-lamp test and chi-square test (or Fisher’s exact test) were per- biomicroscopy, IOP measurements by Goldmann applana- formed to compare the baseline characteristics and tion tonometry, anterior chamber angle examinations with treatment outcomes between the two groups. Logistic re- Posner four-mirror goniolens, specular microscopy, and di- gression analysis was performed to examine the predictive lated fundus exam with FFA. The medical records of the factors for IOP-lowering surgery. Paired t-tests were also patients were reviewed for age, sex, BCVA, preexisting used to evaluate corneal endothelial changes after in- ischemic ocular disorders, the number of topical anti-glau- tracameral bevacizumab injection. A p-value of <0.05 was coma medications, and prior treatments. NVI and NVA considered statistically significant. grade by Weiss and Gold classification were also assessed, which distinguish four stages of neovascularization, ac- cording to the area of new vessels in the iris, anterior Results chamber angle, and the location of PAS [13]. Grading of NVI and NVA was performed by a single glaucoma spe- The baseline characteristics of included patients are cialist (SWP). NVI grading is as follows: grade 1, fine sur- summarized in Table 1. From a total of 34 eyes of 34 pa-

540 JY Ha, et al. Intracameral Bevacizumab in Neovascular Glaucoma

tients who received intracameral bevacizumab, 26 eyes of chamber of the eye examined due to severe corneal edema 26 patients were included in this study. Of the eight eyes or hyphema. Of the 26 included patients, 17 were men and excluded from analyses, four eyes had not been followed- 9 were women. The mean age was 58.3 ± 12.8 years (range, up for 12 months, and four eyes could not have the anterior 42 to 78 years). At the initial visit, BCVA was 1.2 ± 0.8 logMAR, and IOP was 39.8 ± 5.3 mmHg, respectively. An Table 1. Baseline characteristics of patients average of 2.8 ± 0.2 topical anti-glaucoma medications Variable Value were administered. The underlying ophthalmic conditions No. of eyes / patients 26 / 26 causing NVG were PDR (13 eyes, 50%), central retinal Age (yr) 58.30 ± 12.81 vein occlusion (7 eyes, 26%), and ocular ischemic syn- Male / female 17 / 9 drome (6 eyes, 23%). Before receiving intracameral beva- Best-corrected visual acuity (logMAR) 1.22 ± 0.78 cizumab injection, 12 (46%), 7 (26%), and 16 (61%) eyes Intraocular pressure (mmHg) 39.79 ± 5.33 had undergone PRP, pars plana (PPV), and cat- Causes of neovascular glaucoma aract surgery, respectively. Proliferative 13 (50.00) The details of therapeutic intervention in the non-surgi- Central retinal vein occlusion 7 (26.62) cal and surgical group are summarized in Table 2. During Ocular ischemic syndrome 6 (23.08) the entire follow-up period, 19 patients received surgical Topical glaucoma medication 2.75 ± 0.22 treatment. The surgical procedures were as follows: five Previous treatment eyes (26%) underwent trabeculectomy with mitomycin C, 12 eyes (63%) underwent Ahmed valve implantation with Panretinal photocoagulation 12 (46.15) mitomycin C, and two eyes (10%) underwent trans scleral Pars plana vitrectomy 7 (26.92) cyclophotocoagulation. The total number of injections was surgery 16 (61.53) 2.40 ± 1.72 in the non-surgical group and 2.55 ± 2.17 in the Values are presented as number, mean ± standard deviation, or number (%). surgical group (p = 0.495). Five eyes (71%) in the non-sur- logMAR = logarithm of the minimum angle of resolution. gical group and 12 eyes (63%) in the surgical group re-

Table 2. Details of therapeutic intervention in the non-surgical and surgical group Non-surgical group Surgical group p-value No. of eyes 7 (26.92) 19 (73.07) Total injections 2.40 ± 1.72 2.55 ± 2.17 0.495 No. of eyes receiving multiple injections 5 (71.42) 12 (63.16) 0.103 Interval between initial injection and surgery (day) NA 33.61 ± 26.89 IOP at the time of the surgery (mmHg) NA 42.24 ± 8.72 PRP characteristics Complete PRP 6 (85.71) 10 (52.63) 0.021* Power (mW) 352.24 ± 14.75 368.24 ± 13.72 0.354 Laser spot number 1,868.24 ± 363.74 1,452.24 ± 414.75 0.014* Laser spot area 0.034* 1–2 quadrant 1 8 3–4 quadrant 6 11 Surgical treatment Trabeculectomy NA 5 (26.31) Ahmed valve implantation NA 12 (63.16) Trans scleral cyclophotocoagulation NA 2 (10.53) Values are presented as number (%) or mean ± standard deviation. NA = not applicable; IOP = intraocular pressure; PRP = panretinal photocoagulation. *p < 0.05 is considered statistically significant.

541 Korean J Ophthalmol Vol.31, No.6, 2017

quired a repeated injection (p = 0.103). During the fol- mmHg in 22 of 26 eyes. However, 4 of 26 eyes required low-up period, complete PRP was performed in 6 eyes IOP-lowering surgery and IOP was reduced to 12.0 ± 2.8 (85%) in the non-surgical group and 10 eyes (52%) in the mmHg at 1 week after injection. Despite intracameral in- surgical group (p = 0.021). The non-surgical group re- jections and other medical treatments, the number of eyes ceived more spots and quadrant areas than the surgical requiring additional surgical treatment were increased and group (p = 0.014 and p = 0.034, respectively). There was no 14 (53%), 16 (62%), and 19 eyes (73%) had received IOP- difference in laser power. lowering surgeries at 1, 6, and 12 months after injection, The change in IOP after intracameral bevacizumab is respectively. Nevertheless, both groups showed successful shown in Fig. 1. At 1 week, IOP was stabilized to 16.5 ± 3.4 IOP normalization (<20 mmHg) during a 1-year follow-up period.

Total Non-surgical group Surgical group Serial change in NVI and NVA grade in total patients during the follow-up period is shown in Table 3. More than (n = 26) 70 % of eyes were distributed as advanced grade (i.e., 3 or 40 39.79 4) in NVI and NVA at baseline. NVI and NVA rapidly re- 30 gressed after injection. NVI disappeared in 15 eyes (58%) (n = 22) (n = 12) (n = 11) (n = 10) (n = 7) 16.51 17.53 17.52 18.27 18.64 and NVA disappeared in 6 eyes (23%) and eyes with ad- 20 vanced grade in NVI and NVA were reduced to 5 eyes, re- IOP (mmHg) 10 14.55 13.72 13.35 15.01 spectively (18%) (p = 0.01 in NVI and NVA) 1 week after 12.03 (n = 14) (n = 15) (n = 16) (n = 19) (n = 4) injection. This trend continued in both NVI and NVA 1 0 Baseline 1 wk 1 mon 3 mon 6 mon 12 mon month after injection (p = 0.02 and p = 0.04, respectively). Time after initial intracameral bevacizumab injection However, at 3 months post-injection, the effect of in- Fig. 1. The changes in intraocular pressure (IOP) after intracam- tracameral injection was maintained in NVI (p = 0.03) but eral bevacizumab injection. Of 26 eyes, IOP in 22 eyes could not in N VA ( p = 0.07). Intra-rater reliability for NVI and be controlled with injection, but 4 eyes received anti-glaucoma surgery 1 week after injection. At 1 month, 14 eyes received the NVA grading was evaluated by calculation of Cohen’s surgery, and the number increased to 19 eyes at 12 months after kappa coefficient. Coefficient values were 0.86 (95% confi- injection. IOP in the eyes of the non-surgical group was main- tained <20 mmHg after treatment. The surgical group showed dence interval [CI], 0.77 to 0.91) for NVI grading and 0.87 poor response to injection, but, after surgery IOP also stabilized. (95% CI, 0.74 to 0.90) for NVA grading. The kappa values

Table 3. Serial changes for NVI and NVA in total patients during 12-month follow-up Follow-up Grade Baseline 1 wk 1 mon 3 mon 6 mon 12 mon NVI 0 0 (0) 15 (57.7) 8 (30.8) 7 (26.9) 6 (23.1) 6 (23.1) 1 2 (7.7) 3 (11.5) 5 (19.2) 5 (19.2) 4 (15.4) 4 (15.4) 2 3 (11.5) 3 (11.5) 3 (11.5) 4 (15.4) 3 (11.5) 3 (11.5) 3 10 (38.5) 2 (7.7) 5 (19.2) 4 (15.4) 6 (23.1) 7 (26.9) 4 11 (42.3) 3 (11.5) 5 (19.2) 6 (23.1) 7 (26.9) 6 (23.1) p-value* 0.011† 0.024† 0.034† 0.067 0.075 NVA 0 0 (0) 6 (23.1) 5 (19.2) 4 (15.4) 4 (15.4) 3 (11.5) 1 3 (11.5) 9 (34.6) 4 (15.4) 4 (15.4) 3 (11.5) 3 (11.5) 2 4 (15.4). 6 (23.1) 6 (23.1) 5 (19.2) 5 (19.2) 5 (19.2) 3 9 (34.6) 3 (11.5) 5 (19.2) 6 (23.1) 7 (26.9) 7 (26.9) 4 10 (38.5) 2 (7.7) 6 (23.1) 7 (26.9) 7 (26.9) 8 (30.8) p-value* 0.009† 0.041† 0.072 0.089 0.097 Values are presented as number (%). NVI = neovascularization of iris; NVA = neovascularization of anterior chamber angle. *Paired McNemar test. Compared to baseline; †p < 0.05 is considered statistically significant.

542 JY Ha, et al. Intracameral Bevacizumab in Neovascular Glaucoma

for both grades were up to 0.9, which was sufficient to en- 12.51) and complete PRP (p = 0.005; OR, 0.28; 95% CI, 0.09 sure reasonable reliability [14]. to 0.73) were strong predictors for IOP-lowering surgery. Treatment results at 12-month follow-up are summarized Notably, the odds of IOP-lowering surgery increased as in Table 4. BCVA remained relatively stable during the baseline grade of NVA increased and were 7.01 times course of treatment (1.2 ± 0.8 vs. 1.4 ± 0.5 logMAR at higher in patients with grade 4 than in patients with grade baseline and 12-month follow-up, respectively; p = 0.542). 1 (95% CI, 2.65 to 30.67) (Table 5). There was a reduction in IOP and the number of topical Regarding the cytotoxicity of intracameral bevacizumab anti-glaucoma medications after treatment (39.8 ± 5.3 vs. on the corneal endothelium, cell density, coefficient of 1.4 ± 0.5 mmHg, 2.8 ± 0.2 vs. 1.0 ± 0.3 at baseline and variation, and hexagonality of corneal endothelial cells 1 12-month follow-up, respectively; all p < 0.05). Comparing month after injection did not significantly change com- the non-surgical and the surgical groups, there were no pared to baseline in all patients. For exclusion of the effect differences in BCVA or IOP at 12-month follow-up. IOP of surgery on corneal endothelium, the same parameters was stabilized in both groups, but the non-surgical group only in the non-surgical group were analyzed, and there was treated with a greater average number of medications was no obvious corneal endothelial damage at 12-month than the surgical group (1.3 ± 0.5 vs. 0.8 ± 0.3, p < 0.05). In follow-up, despite receiving multiple intracameral bevaci- addition, NVI and NVA were more advanced in the surgi- zumab injections. Comparing baseline values to those at cal group than in the non-surgical group (p = 0.041 and p 12 months in the non-surgery patients, there were no sta- = 0.020, respectively). tistical differences in endothelial cell density (2,238 ± 189 The predictive factors for IOP-lowering surgical treat- vs. 2,198 ± 210, p = 0.120), coefficient of variation (17.4 ± 5.5 ments were analyzed, and the results are shown in Table 5. vs. 18.3 ± 4.9, p = 0.267), and hexagonality (58.2 ± 4.0 vs. Multivariate logistic regression analysis showed that base- 55.0 ± 3.5, p = 0.421) (Table 6). line IOP (p = 0.018; odds ratio [OR], 4.45; 95% CI, 2.79 to

Table 4. Treatment results at 12-month follow-up Total Non-surgery Surgery Variable p-value p-value Baseline 12 mon 12 mon 12 mon No. of eyes 26 26 7 (26.9) 19 (73.1) BCVA (logMAR) 1.2 ± 0.8 1.4 ± 0.5 0.542 1.3 ± 0.7 1.5 ± 0.9 0.579 IOP (mmHg) 39.8 ± 5.3 16.0 ± 4.7 0.001* 18.6 ± 4.3 15.0 ± 5.2 0.246 Topical anti-glaucoma medication 2.8 ± 0.2 1.0 ± 0.3 0.014* 1.3 ± 0.5 0.8 ± 0.3 0.042* NVI grade 0.075 0.041* Grade 0 0 (0) 6 (23.1) 2 (28.6) 4 (21.1) Grade 1 2 (7.7) 4 (15.4) 2 (28.6) 2 (10.5) Grade 2 3 (11.5) 3 (11.5) 1 (14.3) 2 (10.5) Grade 3 10 (38.5) 7 (26.9) 1 (14.3) 6 (31.6) Grade 4 11 (42.3) 6 (23.1) 1 (14.3) 5 (26.3) NVA grade 0.097 0.020† Grade 0 0 (0) 3 (11.5) 3 (42.9) 0 (0) Grade 1 3 (11.5) 3 (11.5) 2 (28.6) 1 (5.3) Grade 2 4 (15.4) 5 (19.2) 2 (28.6) 3 (15.8) Grade 3 9 (34.6) 7 (26.9) 0 (0) 7 (36.8) Grade 4 10 (38.5) 8 (30.8) 0 (0) 8 (42.1) Values are presented as number (%) or mean ± standard deviation. BCVA = best-corrected visual acuity; logMAR = logarithm of the minimum angle of resolution; IOP = intraocular pressure; NVI = neo- vascularization of iris; NVA = neovascularization of anterior chamber angle. *p < 0.05 is considered statistically significant.

543 Korean J Ophthalmol Vol.31, No.6, 2017

Discussion effect of directly targeting VEGF, intraocular bevacizum- ab was established as a standard protocol for NVG treat- The definitive treatment for NVG is the elimination of ment [15]. However, most of the patients who received in- neovascular stimuli by ablation of the ischemic retina, traocular bevacizumab required additional injections, PRP, thereby reducing vasoproliferative factors (e.g., VEGF) or in the long-term, surgical treatment for IOP control and abnormal new vessel formation. Although the effect [12,16-18]. Generally, a drug with intracameral injection of retinal ablation is definitive and long lasting, it often shows shorter half-life and duration of effect than those takes several weeks to manifest [3]. During this period, administered via intravitreal injection. However, intracam- progressive angle closure and damage may eral injection can be performed in the presence of media persist due to sustained VEGF and IOP elevation, and may opacities, has fewer vitreoretinal complications, and shows result in painful vision loss. Because of its rapid and potent better IOP-lowering effects than intravitreal injection in

Table 5. Predictive factors for IOP-lowering surgery Univariate analysis Multivariate analysis Predictive factor Odds ratio (95% CI) p-value Odds ratio (95% CI) p-value Baseline BCVA (logMAR) 2.72 (1.25–45.69) 0.072 Baseline IOP 5.68 (2.21–9.17) 0.001 4.45 (2.79–12.51) 0.018* Baseline NVI grade Grade 1 1 (reference) 1 (reference) Grade 2 1.35 (1.10–1.66) 0.102 Grade 3 1.82 (1.01–7.64) 0.084 Grade 4 2.54 (1.95–5.69) 0.072 Baseline NVA grade Grade 1 1 (reference) 1 (reference) Grade 2 3.83 (2.37–6.19) 0.037 4.01 (2.15–8.21) 0.037* Grade 3 6.67 (1.64–27.09) 0.042 6.42 (1.25–22.97) 0.042* Grade 4 7.18 (1.05–27.62) 0.034 7.01 (2.65–30.67) 0.029* Complete PRP 0.24 (0.11–0.64) 0.025 0.28 (0.09–0.73) 0.005* IOP = intraocular pressure; CI = confidence interval; BCVA = best-corrected visual acuity; logMAR = logarithm of the minimum angle of resolution; NVI = neovascularization of iris; NVA = neovascularization of anterior chamber angle; PRP = panretinal photocoagulation. *p < 0.05 is considered statistically significant.

Table 6. Changes in corneal endothelial parameters after intracameral bevacizumab injection Endothelial cell density Coefficient of variation Hexagonality Total patients Baseline 2,254 ± 241 17.78 ± 4.78 58.68 ± 3.89 1 mon 2,245 ± 223 17.97 ± 5.23 57.32 ± 3.57 Non-surgical patients Baseline 2,238 ± 189 17.42 ± 5.46 58.20 ± 4.01 1 wk 2,214 ± 95 17.28 ± 4.07 57.99 ± 3.69 1 mon 2,207 ± 126 17.65 ± 5.99 58.12 ± 6.07 3 mon 2,217 ± 46 17.53 ± 3.14 57.45 ± 3.69 6 mon 2,206 ± 132 17.80 ± 6.90 56.22 ± 4.01 12 mon 2,198 ± 210 18.28 ± 4.89 55.20 ± 3.50 Values are presented as mean ± standard deviation. p < 0.05 is considered statistically significant (paired t-test).

544 JY Ha, et al. Intracameral Bevacizumab in Neovascular Glaucoma

some reports [8,16]. Therefore, this study investigated the ceived 2.40 ± 1.72 injections, and five eyes (71%) required long-term efficacy and safety of intracameral bevacizum- more than two injections. They also had additional PRP, ab injections and the predictive factors for eventual IOP- and six eyes (86%) had complete PRP. lowering surgery. We identified the predictive factors for IOP-lowering Previous studies have reported the rapid and marked re- surgery after intracameral bevacizumab in NVG patients duction of abnormal new vessels and IOP after intraocular by using multivariate analysis. Patients with higher initial bevacizumab in NVG patients [6,9]. Our study also showed IOP and advanced NVA grade tended to have IOP-lower- a similar beneficial effect in NVG patients. IOP, NVI, and ing surgery. In addition, patients who underwent complete NVA rapidly decreased within 1 week, and the effect con- PRP had less frequent surgical treatments. Most of eyes in tinued to 1 month in NVA, and 3 months in NVI. Howev- the non-surgical group (6 of 7 eyes, 86%) showed more er, despite multiple intracameral bevacizumab injections, frequent complete PRP and showed less NVA than the the number of patients who required IOP-lowering surgery surgical group. This suggests that patients with early stage increased during the follow-up period. Consequently, 19 of NVG responded better to intracameral bevacizumab injec- 26 eyes (73%) had received IOP-lowering surgery by 12 tions and tended to require IOP-lowering surgery to a less- months. Of 19 eyes that received IOP-lowering surgery, the er extent. Nakano et al. [19] recently reported that NVG majority of eyes (15 eyes, 79%) received surgery within patients with retinal vein occlusion had a higher degree of the first 3 months after initial injection. It is very unlikely angle closure and higher IOP, but NVG patients with PDR that the effect of intracameral bevacizumab (ICB) in con- had lower IOP and better visual acuity than other underly- trolling IOP would be maintained over 3 months. In addi- ing conditions. However, we did not find a significant rela- tion, there was a reduction in the number of topical an- tionship between preexisting ocular ischemic disease and ti-glaucoma medications in both groups, but the surgical prognostic factors for IOP-lowering surgery. Previous group was treated with fewer medications than the studies showed various results for corneal endothelial tox- non-surgical group. We believe this difference is mainly icity induced by intracameral bevacizumab injections. owing to the effect of the IOP-lowering surgery. Hosny et al. [20] reported reduced mean endothelial cell Although patients who eventually required IOP-lowering count and hexagonality after intracameral bevacizumab, surgery received a similar number of injections compared but there was no clinically significant corneal edema. Park to the non-surgical group, their IOP remained elevated. In et al. [21] reported that intracameral bevacizumab (up to contrast to NVI, which showed rapid regression and stabi- 2.5 mg/0.1 mL) did not affect endothelial cell viability or lized up to 3 months after injection, NVA in the surgical morphology in rabbit . In our study, we found no group showed less response to injection. Actually, NVA significant corneal endothelial damage in patients through- progressed regardless of treatment in the surgery group. out the follow-up periods. Although corneal endothelial Previous reports have shown various results for NVI and parameters tended to decrease after intracameral bevaci- NVA regression after intraocular bevacizumab injection. zumab injection, the change was not statistically signifi- Bhagat et al. [16] reported that NVA was significantly re- cant. Therefore, our results suggest intracameral bevaci- duced up to 8 weeks after intravitreal or combined intra- zumab injection is a relatively safe procedure for the vitreal and intracameral bevacizumab injections. Yazdani treatment of NVG. et al. [17] reported variable regression of NVA after intra- This study has several limitations. First, this is a retro- vitreal bevacizumab injections. However, Kotecha et al. spective study with a relatively small sample size (26 eyes). [18] reported that NVI was rapidly resolved 1 week after In addition, there was no control group (i.e., non-treated) intravitreal bevacizumab injection, but there was no sig- or other route of injection (e.g., intravitreal or combined nificant reduction in NVA. intracameral and intravitreal) against which to compare Seven eyes could be treated non-surgically in this study. the effect of intracameral injection. Second, reduced ab- IOP, NVI, and NVA rapidly decreased after intracameral normal new vessels and controlled IOP after treatment bevacizumab and stabilized during the 12 months of fol- might be the effect of not only ICB but other treatment low-up. However, a single intracameral injection was not factors (e.g., PRP or IOP-lowering surgery). Actually, PRP enough for IOP and new vessel control. These patients re- was more completely performed in the nonsurgical group

545 Korean J Ophthalmol Vol.31, No.6, 2017

than the surgical group. However, these data suggest that 6. Sasamoto Y, Oshima Y, Miki A, et al. Clinical outcomes ICB in NVG patients, especially in early stage NVG, can and changes in aqueous vascular endothelial growth factor provide a temporary therapeutic window for the comple- levels after intravitreal bevacizumab for iris neovascular- tion of PRP and therefore reduce the risk of IOP-lowering ization and neovascular glaucoma: a retrospective two- surgery. Third, this study measured NVI and NVA only dose comparative study. J Ocul Pharmacol Ther 2012;28: via slit-lamp microscopy and gonioscopy. If we had evalu- 41-8. ated the changes in new vessels with anterior segment flu- 7. Boyd SR, Zachary I, Chakravarthy U, et al. Correlation of orescein angiographic imaging and measured VEGF levels increased vascular endothelial growth factor with neovas- in the aqueous humor after treatment, we could have ex- cularization and permeability in ischemic central vein oc- plained the effect of bevacizumab more objectively. Nev- clusion. Arch Ophthalmol 2002;120:1644-50. ertheless, this is the first investigation of the efficacy and 8. Ehlers JP, Spirn MJ, Lam A, et al. Combination intravitreal safety of intracameral bevacizumab injection in a Korean bevacizumab/panretinal photocoagulation versus panreti- population at 1-year follow-up. nal photocoagulation alone in the treatment of neovascular In conclusion, intracameral bevacizumab rapidly im- glaucoma. Retina 2008;28:696-702. proved NVI, NVA, and IOP in NVG patients. However, 9. Wolf A, von Jagow B, Ulbig M, Haritoglou C. Intracameral the therapeutic effect was transient, and the majority of injection of bevacizumab for the treatment of neovascular patients required multiple injections, additional laser treat- glaucoma. Ophthalmologica 2011;226:51- 6. ment, and IOP-lowering surgery. The predictive factors for 10. Krohne TU, Eter N, Holz FG, Meyer CH. Intraocular phar- IOP-lowering surgery were baseline IOP, NVA grade, and macokinetics of bevacizumab after a single intravitreal in- complete PRP. jection in humans. Am J Ophthalmol 2008;146:508-12. 11. Yazdani S, Hendi K, Pakravan M, et al. Intravitreal bevaci- zumab for neovascular glaucoma: a randomized controlled Conflict of Interest trial. J Glaucoma 2009;18:632-7. 12. Wakabayashi T, Oshima Y, Sakaguchi H, et al. Intravitreal No potential conflict of interest relevant to this article bevacizumab to treat iris neovascularization and neovascu- was reported. lar glaucoma secondary to ischemic retinal diseases in 41 consecutive cases. Ophthalmology 2008;115:1571-80. 13. Weiss DI, Gold D. Neofibrovascularization of iris and ante- References rior chamber angle: a clinical classification. Ann Ophthal- mol 1978;10:488-91. 1. Shazly TA, Latina MA. Neovascular glaucoma: etiology, 14. Fleiss JL. Statistical methods for rates and proportions. diagnosis and prognosis. Semin Ophthalmol 2009;24:113-21. 2nd ed. New York: John Wiley and Sons; 1981. p. 212-3. 2. Sivak-Callcott JA, O’Day DM, Gass JD, Tsai JC. Evidence- 15. Sugimoto Y, Mochizuki H, Okumichi H, et al. Effect of in- based recommendations for the diagnosis and treatment of travitreal bevacizumab on iris vessels in neovascular glau- neovascular glaucoma. Ophthalmology 2001;108:1767-76. coma patients. Graefes Arch Clin Exp Ophthalmol 2010; 3. Doft BH, Blankenship G. Retinopathy risk factor regres- 248:1601-9. sion after laser panretinal photocoagulation for prolifera- 16. Bhagat PR, Agrawal KU, Tandel D. Study of the effect of tive diabetic retinopathy. Ophthalmology 1984;91:1453-7. injection bevacizumab through various routes in neovascu- 4. Aiello LP, Avery RL, Arrigg PG, et al. Vascular endothelial lar glaucoma. J Curr Glaucoma Pract 2016;10:39-48. growth factor in ocular fluid of patients with diabetic reti- 17. Yazdani S, Hendi K, Pakravan M. Intravitreal bevacizum- nopathy and other retinal disorders. N Engl J Med 1994; ab (Avastin) injection for neovascular glaucoma. J Glauco- 331:1480-7. ma 2007;16:437-9. 5. Tripathi RC, Li J, Tripathi BJ, et al. Increased level of vas- 18. Kotecha A, Spratt A, Ogunbowale L, et al. Intravitreal be- cular endothelial growth factor in aqueous humor of pa- vacizumab in refractory neovascular glaucoma: a prospec- tients with neovascular glaucoma. Ophthalmology 1998; tive, observational case series. Arch Ophthalmol 2011;129: 105:232-7. 145-50.

546 JY Ha, et al. Intracameral Bevacizumab in Neovascular Glaucoma

19. Nakano S, Nakamuro T, Yokoyama K, et al. Prognostic cells: an in vivo evaluation. J Ocul Pharmacol Ther 2009; factor analysis of intraocular pressure with neovascular 25:513-7. glaucoma. J Ophthalmol 2016;2016:1205895. 21. Park HY, Kim SJ, Lee HB, et al. Effect of intracameral be- 20. Hosny MH, Zayed MA, Shalaby AM, Eissa IM. Effect of vacizumab injection on corneal endothelium in rabbits. intracameral bevacizumab injection on corneal endothelial Cornea 2008;27:1151-5.

547